Detecting the accurate moving objects in indoor stadium using you only look once algorithm compared with ResNet50
Shaik Khasim Saida, V. Parthipan · 2025
The aim of this study is to use You Only Look Once to follow players and their movements across indoor stadiums, and to compare the results with ResNet 50 to enhance accuracy. This study is divided into two groups: You Only Look Once and ResNet 50. Sample size is 10 for each group, calculated using ClinCalc software with a 95% confidence interval and a 0.05 alpha value. The motion detection dataset with a size of 1500 images was considered from Kaggle.com. It was discovered that the proposed method achieved 95.680% accuracy when compared to the current system, which only had an accuracy of 93.270%. Evaluating independent sample T-test, it was proved that YOLO was much more accurate than and SMD models, with a statistical difference between the value of p=0.000 (Independent sample t-test p<0.05). The You Only Look Once method is noticeably superior to the ResNet50 algorithm at detecting moving objects in indoor stadiums according to experimental results.